Minimization in PyROOT

yburkard · February 2, 2023, 2:59pm

Dear Users,

I am fairly new to ROOT and wanted to minimise a function using the PyROOT framework, but don’t really know where to begin. I have stumped upon the TMinuit and Math.Minimizer classes, but there are very few tutorials available (or at least I haven’t found them), and I’m not sure if one has better performance / is more commonly used today. Any hints and recommendations would be appreciated!

Wile_E_Coyote · February 2, 2023, 4:42pm

${ROOTSYS}/tutorials/fit/

moneta · February 8, 2023, 11:56am

Hi,
The recommended way to minimise a function is to use the ROOT.Math.Minimizer class. There is this tutorial, ROOT: tutorials/fit/NumericalMinimization.C File Reference, but it is in C++. If you have issues translating it in Python, please let me know, I can provide that example also in Python

Lorenzo

yburkard · February 8, 2023, 1:28pm

Dear @moneta, thanks for your hints. An example in python would be in fact very useful, as I am not very acquainted with C++. Many thanks in advance!

moneta · February 9, 2023, 5:16pm

Hi,
Here is the Python tutorial for minimising a function:

github.com

lmoneta/root/blob/add_minim_tutorial/tutorials/fit/NumericalMinimization.py

#  \file
#  \ingroup tutorial_fit
#  \notebook -nodraw
#  Example on how to use the new Minimizer class in ROOT
#   Show usage with all the possible minimizers.
#  Minimize the Rosenbrock function (a 2D -function)
#  This example is described also in
#  http://root.cern.ch/drupal/content/numerical-minimization#multidim_minim
#  input : minimizer name + algorithm name
#  randomSeed: = <0 : fixed value: 0 random with seed 0; >0 random with given seed
#
#  \macro_code
#
#  \author Lorenzo Moneta

import ROOT

def RosenBrock( vecx ):
   x = vecx[0]
   y = vecx[1]

This file has been truncated. show original

It will be soon merged into ROOT master

Lorenzo

yburkard · February 9, 2023, 5:34pm

Thank you very very much for the tutorial!

yburkard · February 10, 2023, 2:01pm

Hi @moneta, I was looking at your code, and wanted to point out two possible typos:
(1) In line 37 I think you mean MinimizerName
(2) In line 76 you perhaps mean if ( ret and abs(minimizer.MinValue()-f(xs)) < 1.E-4 ) :, so that you take the difference between the two minimum values instead of the values themselves, no?

moneta · February 10, 2023, 3:08pm

Thank you very much for the corrections. I will include in the open PR for the tutorial,

Cheers

Lorenzo

system · February 24, 2023, 3:08pm

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.